Research And Implementation Of Radio And Tv Program Recommendation Platform Based On Microservices

With the advent of the Internet + era,the amount of information has exploded,making it difficult for users to quickly obtain the information they really need when facing massive amounts of information,but instead reducing the utilization of information,which causes information overload problem.Correspondingly in multimedia,because of the explosion of multimedia data,users have more choices when watching multimedia resources such as movies and TV,and cannot quickly select the multimedia resources they are interested in.A very potential solution to the problem of information overload is a recommendation system.The original intention of the recommendation system design is to collect the user's behavior,combine the material data of the material pool,and model the behavior data and material data to match the similarity between them,so as to recommend materials that may be of interest to users.The traditional monolithic development model assumes that the system is highly coupled,which is neither easy to develop nor easy to maintain later in the system.The best solution to these problems is the development of microservices.In the microservices framework,each functional component has high cohesion,low coupling between services,and data transmission between services depends on network communication,which is very friendly to system development and maintenance.This thesis will focus on the design and implementation of a micro-service-based program recommendation platform for the broadcast and television industry.The purpose is to enable operators in the radio and television industry to use the platform to configure a little according to their business needs and follow the linear process of the recommendation system.Thus,a suitable broadcasting and television recommendation system can be quickly launched.My main work is to split the system's sub-services according to the recommended system operation process,from the recommended scenario configuration,data source management to task configuration to the final recommendation result acquisition,the different processes are independently developed in sub-modules,and the smooth application of micro services Development model.Different modules perform specific functions in a certain stage of the recommendation system process.Finally,the platform implements a complete solution for program recommendation for the broadcast industry.At the same time,the task configuration module uses a layered idea,which is divided into offline tasks,near-line tasks,and online tasks.The results are updated to ensure that they are not out of date.When the recommendation platform proposed in this thesis is actually running,customers can use it to quickly launch the radio and television program recommendation service,the number of user requests rises rapidly,and the user click-through rate(UCTR)and subscription rate increase significantly.At the same time,the platform supports a variety of recommended industries and recommended scenarios,and can be flexibly updated according to business needs.In terms of platform performance,relying on DevOps practice can meet the performance requirements in high concurrency and high availability scenarios.